Anglo American Mining: Geoscience ML Platform

The Project

Embedded as a front-end developer within a global development programme (8 projects, 70+ developers), building a machine learning tool to predict copper porphyry deposit fertility from core samples and laboratory results. My immediate team was distributed across Perth, London and South Africa. The platform gave geoscientists a visual interface to interact with ML model outputs that would otherwise live entirely in notebooks and scripts.

My Role

I built React/TypeScript front-end components against an Azure-hosted component artifact library, working directly alongside geoscientists and data scientists to translate complex ML model outputs into usable interfaces. The back-end was Python and .NET, with Jupyter Notebooks used throughout for data exploration and model experimentation.

I ran UX design workshops and user testing sessions with geoscientists, users whose domain expertise dwarfed anything I could bring technically, but whose tooling was built by ML engineers, not designers. The gap between what the models could do and what the geoscientists could practically interrogate was where UX added the most value. This was a clear example of the technical translator role: bridging data scientists, ML engineers and domain experts who each spoke different languages about the same problem.

Working at Scale

This was my first experience working within a SAFe Agile framework at genuine scale, coordinating across multiple project teams, managing dependencies, and navigating the ceremony and structure that comes with a 70+ person programme. Docker Desktop and Windows Subsystem for Linux (WSL) containerised development environments kept the local toolchain consistent across the distributed team.

Why This Matters

This project combined three things that rarely appear together: enterprise-scale React/TypeScript development, direct collaboration with data scientists on ML tooling, and UX design for a domain (geoscience) where the users are deep subject matter experts. The speed at which I picked up completely unfamiliar technologies (Python), ML/AI concepts, and geoscience domain knowledge, and delivered real value, is one of the strongest examples of rapid technology uptake in my career. It was my first enterprise AI/ML project, and it established the pattern (design the interface between human expertise and machine capability) that later shaped AI adoption work at Insight.

Hard Skills

React, TypeScript, Python, .NET, Azure, Docker, WSL, Jupyter Notebooks, SAFe Agile, component artifact libraries

Soft Skills

Technical translation (data scientists ↔ ML engineers ↔ geoscientists), rapid technology uptake (Python, ML/AI, geoscience domain), cross-disciplinary collaboration, UX workshop facilitation, user testing, working at scale in large distributed teams (Perth/London/South Africa)